AnchorSIPS: Synthetic Dataset for Psychosis-Risk Assessment

Researchers have introduced AnchorSIPS, a synthetic dataset of 10,000 structured psychosis-risk interviews designed to overcome data-access barriers in AI research. The dataset, detailed in a paper on arXiv (2608.12329), models each interview on the Mini-SIPS, a clinician-administered tool for assessing psychosis risk. It includes patient history, 24 symptom questions, follow-up evidence for endorsed items, and decisions on delusion-like, hallucination-like, and disorganized communication symptoms. It also excludes clear psychotic-level symptoms and provides a final diagnosis of attenuated psychosis syndrome (APS), a high-risk state. The APS diagnosis is not standalone but depends on earlier endorsements and supporting details. This resource aims to facilitate progress in AI for psychosis-risk assessment by providing a shareable, privacy-compliant dataset.